Not related to but, as you can see a trend in the pings over time on the graph it reminds me that even in a very granular benchmark it can be very helpful to plot the results first before trying to process them.
For example, we sometimes blindly calculate the mean whereas the data may be such that we end up giving a value the benchmark will never reach.
Well pings will have a bit of a weird distribution. If you did a histogram of your ping for a bit you would see most values sitting near some "floor" of X ms and then a long right-sided tail of values much larger than X ms. It seems to look sorta like a gamma distribution from a quick test of my wi-fi ping. The mean would sit somewhere to the right of the largest mass of ping times. If you wanted to know where most pings were you could look at the median (which will sit right where most pings are). If you were interested in the really large values of X ms then you could compare the mean and median to find out just how much those slow ping responses are dragging your mean away from the median.
For example, we sometimes blindly calculate the mean whereas the data may be such that we end up giving a value the benchmark will never reach.